AI toolkit · free download
You have a meeting about a service you have never opened
Paste the readme and the commit log. Get one page, and three questions worth asking in the room.
The service catch-up pack is a free prompt pack that builds a one-page model of an unfamiliar service. Paste the readme, the entry-point files and the commit log into any AI assistant. You get one page: what it does, who breaks when it does, where it breaks, who owns it, and three questions to ask.
Who this is for
- Engineering managers who just inherited systems somebody else built.
- Consultants and contractors joining a codebase this week.
- Anyone with a meeting in twenty minutes about a service they have never opened.
Before you download. This pack runs in your environment, on your data. Nothing is sent to Perform, and Perform receives no data from it at any point. It is provided free and as is, with no warranty of any kind. It is not legal, tax, security or accounting advice, not an audit and not an attestation, and no professional relationship is created by downloading or using it. You are responsible for reviewing it before you run it and for any decision you make with its output. Licensed under Apache-2.0, which includes a disclaimer of warranties and a limitation of liability.
Anything you point it at should be treated as untrusted input, because a document, a repository or a chat thread can carry instructions aimed at a model.
Version 1.0, 28 August 2026. Security contact: privacy@totalperform.com, acknowledged within five business days.
A markdown file. Paste it into any assistant. Nothing to install.
The same pack, installed once and there every time.
The problem
- The readme describes a system from eighteen months ago.
- The architecture note mentions a queue the code no longer uses.
- The file that matters most is the one nobody wrote a comment in.
- What earns you credibility in the room is not in any single input.
What it does about it
- One page, thirteen fields: filled as you read, so nothing gets drafted twice.
- Cross-reads incidents against commits: did a fix ship, did it hold, and who did the fixing.
- Finds the single point of failure: when one name is on every incident fix, you have found the person to ask.
- Ranks damage before likelihood: a rare failure that loses money outranks a weekly one that recovers.
- Hands you three questions: each traceable to a line you actually read.
Two minutes to first output
- Get the log with real dates and paths:
git log --date=short --name-only -50 - Paste the pack, the readme, the entry points, the log and the alert config.
- Read the ownership field and the cross-read first.
What it will not do
- Produce the main path from a route table alone. It marks that field inferred.
- Line up relative commit dates against an incident. It states which readings they permit.
- Assume an incident with no commit after it was never fixed. It checks your log window first.
| What you paste | Readme, directory listing, entry points and handler bodies, commit log, incidents, alert config |
|---|---|
| What comes back | A thirteen-field page, a ranked list of where it breaks, three questions |
| Time | Under two minutes |
| Where it runs | Your assistant, on your machine |
| Tested against | A real-shape fee service with a stale readme and a silent wrong-money path. Four rounds of practitioner testing. |
30 minutes with the engineer who would do the work.